Returns and Order Flow Imbalances: Intraday Dynamics and Macroeconomic News Effects

ArXiv ID: 2508.06788 “View on arXiv”

Authors: Makoto Takahashi

Abstract

We study the interaction between returns and order flow imbalances in the S&P 500 E-mini futures market using a structural VAR model identified through heteroskedasticity. The model is estimated at one-second frequency for each 15-minute interval, capturing both intraday variation and endogeneity due to time aggregation. We find that macroeconomic news announcements sharply reshape price-flow dynamics: price impact rises, flow impact declines, return volatility spikes, and flow volatility falls. Pooling across days, both price and flow impacts are significant at the one-second horizon, with estimates broadly consistent with stylized limit-order-book predictions. Impulse responses indicate that shocks dissipate almost entirely within a second. Structural parameters and volatilities also exhibit pronounced intraday variation tied to liquidity, trading intensity, and spreads. These results provide new evidence on high-frequency price formation and liquidity, highlighting the role of public information and order submission in shaping market quality.

Keywords: Structural VAR, Order Flow Imbalance, Price Impact, High-Frequency Data, Limit Order Book, Futures

Complexity vs Empirical Score

• Math Complexity: 8.5/10
• Empirical Rigor: 9.0/10
• Quadrant: Holy Grail
• Why: The paper employs advanced econometric techniques (structural VAR with heteroskedasticity identification) on high-frequency data, involving complex modeling and identification. It is highly empirical, using massive real-world datasets (one-second data over 1,490 days) with detailed implementation of data cleaning, variable construction, and statistical inference.

  flowchart TD
    A["Research Goal:<br>Intraday Dynamics of Returns<br>vs Order Flow Imbalances"] --> B["Methodology:<br>Structural VAR<br>Identified via Heteroskedasticity"]
    B --> C["Data:<br>S&P 500 E-mini Futures<br>1-Second Frequency"]
    C --> D["Estimation:<br>15-Minute Intervals<br>across Trading Day"]
    D --> E{"Key Findings"}
    E --> F["Macro News Spikes:<br>Price Impact ↑ | Flow Impact ↓"]
    E --> G["Dynamics:<br>Shocks fade within 1s<br>Price/Flow Impact significant"]
    E --> H["Intraday Variation:<br>Tied to Liquidity & Spreads"]